Actually this is not so. Phenomena that repeat themselves are not intrinsic to science.
You may possibly be getting confused about what "repeatability" (or reproducibility) of observations means. It means that different observers can repeat
the observation and get consistent results - basically a way of minimising the subjective element in any human observations of nature.
ToE makes predictions that are regularly shown correct, as any good scientific theory should.
For instance, it generally predicts correctly what degrees of relative similarity in DNA there should be between organisms, from the Phylogenetic Tree, a construction that is one product of the theory.
And it predicts what transitional fossils we should expect to find, and in rocks of what age, and we find them.
(Different observers generally have little trouble agreeing on degrees of DNA similarity, or on what the fossils are, so the need for repeatability poses no problem for the ToE.)
So, regardless of what you may have read in a creationist tract, it is quite wrong to imagine that it is qualitatively different from a theory like, say, the part of chemistry that accounts for the Periodic Table. That too makes predictions of what properties and behaviour we should expect from the elements and it too is fairly successful in doing what it does.
In fact it is the
departures from the predictions that make for the most interesting chemistry. I have no doubt that the same is true of the theory of evolution.